Since September, commuters using Shanghai’s Number 9 bus route have had a new way of catching a ride. Rather than stand at a designated stop, they open a smartphone app and book a ride to wherever they're going. The service, provided by Alibaba, takes those reservations into account and calculates where the bus should go, using the company’s artificial intelligence to customize the route. The idea is to boost ridership—and curb traffic—by making public transit more convenient.
Shanghai is just the latest city to give this sort of scheme a try. From Helsinki, Finland, to Sydney, cities around the world have spent the past few years trying to implement AI-fueled, on-demand bus services. Few have succeeded.
Earlier this year, Singapore decided against renewing a pilot for on-demand buses. In Germany, microtransit company CleverShuttle—which bills itself as more of a ride-pooling service than a bus—pulled out of three of the eight cities it was operating in, citing economic and bureaucratic hurdles. In a pilot project with shared rides company Via, bringing underserved residents to public transit nodes, Los Angeles Metro is spending $14.50 per trip—twice what it spends on a regular bus trip.
On-demand buses have been a thing for decades. Public transit agencies often call them demand-responsive buses, and deploy them to serve users who lack easy access to standard routes because they live especially far away, or may have special needs. Because they reach relatively few people, they’re expensive to operate. They’re inefficient too, often making riders wait undetermined amounts of time for a ride. So cities must strike a balance between making public transit accessible to the largest number of residents, and meeting their budget goals.
Projects like the one in Shanghai represent a new kind of effort. The new, tech-powered services—sometimes called microtransit, because they use small vehicles—claim to make those routes cost-effective and attractive by pairing transit data with the convenience of a smartphone app. The goal is to help transit agencies reach currently underserved populations, such as people who need a ride home from the train station, night riders, or urbanites who’d like to ditch the car but don’t want to use public transit.
According to the tech companies pushing this solution, making on-demand busing work is a matter of crunching vast amounts of transit data, now made available by location tracking, and using algorithms to create custom shared routes. Data will help agencies reroute buses in real time based on factors like user demand and congestion, says Amos Haggiag, CEO of Optibus, whose software helps cities plan and manage bus routes, both on-demand and fixed. “I do see mass transit, even the large buses, as much more dynamic.” Many of those companies, including Uber, think all buses, not just those in low-ridership areas, should run on demand.
Reality, though, adds complications. Not everyone who needs to get around has access to an app. Smartphone ownership remains vastly unequal among countries, and between income and age groups. The cost of data is still cited as a major barrier to smartphone use around the world. And even those who do have phones may not want to rely on them to get to work. When I point out that my smartphone shuts down when the weather gets too cold in winter, Haggiag says my situation is “extreme.” I live in Montreal, along with 1.75 million other people.
Tech companies and planners often make decisions without considering the needs of people who are not like them. A pilot project in St. Petersburg, Florida, that let residents use Uber to connect to bus stops faced low adoption rates. The local transit authority realized residents, many of whom were low-income, didn’t know how to use Uber. They needed help on how to use the app, a planner told WIRED in 2017. Elsewhere, “smart city” initiatives have been called out for their lack of inclusivity.
The problem is “elite projection,” according to public transit consultant Jarrett Walker—“the mistake of not realizing that you are in the minority by virtue of being elite, which means that you may be in love with something that doesn't actually scale to the whole population.” That helps explain why Uber, which has undoubtedly improved the quality of life for those who can afford it, is used by many microtransit companies as a baseline for what a transit experience should look like: available on demand and ultra-convenient. “All of these companies have enormous amounts of venture capital, and therefore an enormous ability to shape the conversation in a way that serves their interests,” Walker says.
If you want to know what happens when transit becomes fully Uberized, look to Innisfil. The small city in Ontario didn’t have enough cash to set up a fully fledged transit system, so it partnered with Uber to offer subsidized rides. The scheme has proven so popular that the municipality has had to cap rides and increase fares to keep it going. “They've had too much demand,” says Moaz Ahmad, a Toronto-based public transit consultant. “When anything grows, it will find plateaus, it will find barriers that have to be surmounted.”
If the success of a transit system depends on the number of people who use it, then successful public transit is everything but on-demand transportation. Cities have been able to boost bus ridership by making improvements not on their ability to reach users where they live, but by increasing frequency, speed, and capacity on the busiest routes. They’ve done this by removing stops or spacing them out, building dedicated bus lanes, and decreasing the time it takes for people to get on and off the bus, for instance by letting people board through rear doors. Ridership and on-time performance soared on Manhattan’s 14th Street bus line after efficiency measures were introduced starting in 2018, while travel times dropped 38 percent.
New tech could deliver real value, though, on a transit system’s most important routes. “People are always talking about first mile, last mile, but no one is dealing with all the miles in between,” says Optibus’ Haggiag. Where many transit authorities still rely on spreadsheets and analog planning tools to design bus routes, the Optibus software can input unlimited data sources to calculate optimal routes and schedules. The city of Herzliya, Israel, saw its bus ridership double after redesigning its bus map and increasing frequency, Haggiag adds.
Time and again, cities have found that making those main corridors more reliable increases incentives for people to find a way to reach bus stops, whether that’s by riding an Uber, a scooter, or a bike. Most often, though, that’s by walking, which is easier said than done. Denver has drafted a plan to improve walkability after finding that 39 percent of sidewalks within a mile of bus stops are either nonexistent or too narrow. Portland, Oregon, is improving access to transit stops as part of its Vision Zero plan to eliminate traffic deaths, because 32 percent of stops have been found to be unsafe for pedestrians. Sometimes the best way to get people on the bus involves no technology at all.
It may be that on-demand buses or microtransit may be the best solution for a city’s needs, but in most cases the answer starts with good governance. “When you are responding to a marketing pitch, you are not thinking,” Walker says. “Thinking begins by having a moral conversation about your goals, about what kind of city you want, and about what you want life in your city to be.”